X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=picoclvr.py;h=774ae3b6af3bd45da09713fa249ab611ece8e580;hb=b5efc396f45c23b7de0fe11f618731ac2b900d99;hp=601bdf7dbe7bce3291b7b7e07cecc1b9587e94ca;hpb=119d5e84350bcab97c06a5c30227a294ebadb3c3;p=mygpt.git

diff --git a/picoclvr.py b/picoclvr.py
index 601bdf7..774ae3b 100755
--- a/picoclvr.py
+++ b/picoclvr.py
@@ -71,8 +71,29 @@ color_tokens = dict( [ (n, c) for n, c in zip(color_names, colors) ] )
 
 ######################################################################
 
+def all_properties(height, width, nb_squares, square_i, square_j, square_c):
+    s = [ ]
+
+    for r, c in [ (k, color_names[square_c[k]]) for k in range(nb_squares) ]:
+        s += [ f'there is {c}' ]
+
+        if square_i[r] >= height - height//3: s += [ f'{c} bottom' ]
+        if square_i[r] < height//3: s += [ f'{c} top' ]
+        if square_j[r] >= width - width//3: s += [ f'{c} right' ]
+        if square_j[r] < width//3: s += [ f'{c} left' ]
+
+        for t, d in [ (k, color_names[square_c[k]]) for k in range(nb_squares) ]:
+            if square_i[r] > square_i[t]: s += [ f'{c} below {d}' ]
+            if square_i[r] < square_i[t]: s += [ f'{c} above {d}' ]
+            if square_j[r] > square_j[t]: s += [ f'{c} right of {d}' ]
+            if square_j[r] < square_j[t]: s += [ f'{c} left of {d}' ]
+
+    return s
+
+######################################################################
+
 def generate(nb, height = 6, width = 8,
-             max_nb_squares = 5, max_nb_statements = 10,
+             max_nb_squares = 5, max_nb_properties = 10,
              many_colors = False):
 
     nb_colors =  len(color_tokens) - 1 if many_colors else max_nb_squares
@@ -91,28 +112,14 @@ def generate(nb, height = 6, width = 8,
         img = [ 0 ] * height * width
         for k in range(nb_squares): img[square_position[k]] = square_c[k]
 
-        # generates all the true relations
+        # generates all the true properties
 
-        s = [ ]
+        s = all_properties(height, width, nb_squares, square_i, square_j, square_c)
 
-        for r, c in [ (k, color_names[square_c[k]]) for k in range(nb_squares) ]:
-            s += [ f'there is {c}' ]
+        # pick at most max_nb_properties at random
 
-            if square_i[r] >= height - height//3: s += [ f'{c} bottom' ]
-            if square_i[r] < height//3: s += [ f'{c} top' ]
-            if square_j[r] >= width - width//3: s += [ f'{c} right' ]
-            if square_j[r] < width//3: s += [ f'{c} left' ]
-
-            for t, d in [ (k, color_names[square_c[k]]) for k in range(nb_squares) ]:
-                if square_i[r] > square_i[t]: s += [ f'{c} below {d}' ]
-                if square_i[r] < square_i[t]: s += [ f'{c} above {d}' ]
-                if square_j[r] > square_j[t]: s += [ f'{c} right of {d}' ]
-                if square_j[r] < square_j[t]: s += [ f'{c} left of {d}' ]
-
-        # pick at most max_nb_statements at random
-
-        nb_statements = torch.randint(max_nb_statements, (1,)) + 1
-        s = ' <sep> '.join([ s[k] for k in torch.randperm(len(s))[:nb_statements] ] )
+        nb_properties = torch.randint(max_nb_properties, (1,)) + 1
+        s = ' <sep> '.join([ s[k] for k in torch.randperm(len(s))[:nb_properties] ] )
         s += ' <img> ' + ' '.join([ f'{color_names[n]}' for n in img ])
 
         descr += [ s ]
@@ -123,23 +130,22 @@ def generate(nb, height = 6, width = 8,
 
 def descr2img(descr, height = 6, width = 8):
 
+    if type(descr) == list:
+        return torch.cat([ descr2img(d) for d in descr ], 0)
+
     def token2color(t):
         try:
             return color_tokens[t]
         except KeyError:
             return [ 128, 128, 128 ]
 
-    def img_descr(x):
-        u = x.split('<img>', 1)
-        return u[1] if len(u) > 1 else ''
-
-    img = torch.full((len(descr), 3, height, width), 255)
-    d = [ img_descr(x) for x in descr ]
-    d = [ u.strip().split(' ')[:height * width] for u in d ]
-    d = [ u + [ '<unk>' ] * (height * width - len(u)) for u in d ]
-    d = [ [ token2color(t) for t in u ] for u in d ]
-    img = torch.tensor(d).permute(0, 2, 1)
-    img = img.reshape(img.size(0), 3, height, width)
+    d = descr.split('<img>', 1)
+    d = d[-1] if len(d) > 1 else ''
+    d = d.strip().split(' ')[:height * width]
+    d = d + [ '<unk>' ] * (height * width - len(d))
+    d = [ token2color(t) for t in d ]
+    img = torch.tensor(d).permute(1, 0)
+    img = img.reshape(1, 3, height, width)
 
     return img
 
@@ -147,13 +153,12 @@ def descr2img(descr, height = 6, width = 8):
 
 if __name__ == '__main__':
     descr = generate(nb = 5)
-    for d in descr:
-        print(d)
-        print()
 
-    img = descr2img(descr)
-    print(img.size())
+    with open('picoclvr_example.txt', 'w') as f:
+        for d in descr:
+            f.write(f'{d}\n\n')
 
+    img = descr2img(descr)
     torchvision.utils.save_image(img / 255.,
                                  'picoclvr_example.png', nrow = 16, pad_value = 0.8)